Certain computing devices have speech recognition functionality, e.g., voice-activated devices, and may provide appropriate responses to a user's utterance. Certain of these computing devices with speech recognition functionality are shared devices that have multiple users interacting with the shared device. In addition, a particular user may interact with several different devices that have speech recognition functionality. In each of these instances, the computing device with speech recognition functionality can have difficulty identifying the particular user that is currently interacting with the computing device. Accordingly, the user's interaction with the computing device often yields undesirable results.